Josef Drexl * Refusal to Deal. Answers to the Questionnaire of the ICN Unilateral Conduct Working Group

Transcription

1 Josef Drexl * Refusal to Deal Answers to the Questionnaire of the ICN Unilateral Conduct Working Group As a non-governmental agent (NGA) and a member of the ICN Unilateral Conduct Working Group, I hereby provide my answers to the questionnaire on refusal to deal. I. Concept of Refusal to Deal The answers are based on the term of refusal do deal as defined by the questionnaire, i.e. as the unconditional refusal by a dominant firm (or a firm with substantial market power) to deal with a rival. The questionnaire further explains that this typically occurs when a firm refuses to sell an input to a company with which it competes (or potentially competes) in a downstream market. In this sense, a refusal to deal also includes the case of a refusal to license intellectual property (IP) rights or to grant access to an essential facility. A refusal to deal may also take the form of a constructive refusal to deal which is characterized by the dominant firm s offering to supply its rival on unreasonable terms (e.g., extremely high prices, degraded service, or reduced technical interoperability). A constructive refusal to deal may be accomplished through a so-called margin-squeeze, which occurs when a dominant firm charges a price for an input in an upstream market, which, compared to the price it charges for the final good using the input in the downstream market, does not allow a rival on the downstream market to compete. Neither the questionnaire nor the following answers to it cover conditional refusals to deal. In such a case, the supply of the relevant product is conditioned on the rival s accepting limitations on its conduct, such as certain tying, bundling, or exclusivity arrangements. The ICN Unilateral Conduct Working Group has dealt with such practices in the Report on Tying and Bundled Discounting (June 2009) and the Report on Exclusive Dealing (April 2008). 1 II. Purpose of These Answers The questionnaire is principally addressed to the different agencies as ICN Members and seeks to produce information on varies jurisdictions. Yet it was also made clear that the NGAs should not double the work of their national agencies but should feel free to present their own view on the issues at stake. Accordingly, the answers take a much broader policy approach to the questions. References to individual jurisdictions are made mostly with the aim of illustrating the substantive arguments. Most examples are taken from European and German law on the one hand and U.S. American law on the other hand. * Dr. iur. (Munich), LL.M. (Berkeley). Director of the Max Planck Institute for Intellectual Property, Competition and Tax Law, Munich and Honorary Professor of Law at the University of Munich. 1 See 1

2 III. General Legal Framework 1. Does your jurisdiction recognize a refusal to deal as a possible violation of your antitrust law? If so, is the term refusal to deal used in a manner different from the definition in the introductory paragraphs above? Please explain. Refusals to deal are recognized by some jurisdictions as a potential violation of competition laws. In the framework of rules on unilateral conduct, a refusal to deal can be considered as one of the cases of non price-related conduct for which illegality is most difficult to be assessed. In the market economy, also dominant firms should in principle be free to choose with whom they prefer to deal and at what terms. As in all cases of unilateral conduct, the challenge consists in distinguishing anticompetitive conduct from competition on the merits. Hence, the question is not whether a refusal to deal is considered illegal but which conditions need to be met to make an individual refusal to deal unlawful. A first step to limiting the rules and principles on refusal to deal may consist in defining the very concept of a refusal to deal more narrowly than just as a refusal to deal with a (potential) customer. The questionnaire seems to take such an approach by limiting refusals to deal to refusals to deal with a rival. Yet such a narrow definition would exclude many cases of potentially illegal refusals to deal in which the dominant firm and the costumer are not active in the same market. Under European law, United Brands, one of the first and most prominent refusal-to-deal cases was of such a nature. 2 In this case, United Brands tried to punish and discipline Olson, a Danish ripener, for having cooperated with Dole in an advertising campaign by terminating supplies of bananas. In this case, Olson was not a rival of United Brands. Rather, the economic target of the refusal to supply was Dole, United Brands competitor, who was in need of access to the ripening facilities controlled by United Brands. This example demonstrates that the cases to be covered in the following should not be limited to refusal to deal with a rival. It is therefore suggested to define a refusal to deal more openly in the sense of a refusal to deal with a (potential) customer. As pointed out before, this would not mean that every refusal to deal with a customer will have to be considered as illegal. This is rather a matter of the additional requirements of illegality, which still need to be defined. In line with this broad definition, it is recommended to leave the theory of harm, which is required in order to distinguish legal from illegal conduct, to the conditions for illegality. This seems important in order to avoid false negatives. Whereas it is true that refusal-todeal cases are nowadays mainly discussed when vertically integrated firms use their market power in the upstream market as a leverage to exclude competitors from a downstream market, United Brands demonstrates that these are not the only relevant cases. The scenario, in which a dominant firm tries to extend market power to a downstream market by refusing to deal with a rival, is not just a description of a potential illegal refusal to deal; it already alludes to a very specific theory of harm. This is especially true for the so-called essential facilities doctrine. Equating an essential facilities scenario with an illegal refusal to deal may easily produce two sorts of fallacies. On the one hand, there is a risk of false negatives if enforcers argue that a refusal to license can only be considered as illegal if the dominant firm refuses to provide an essential input or to grant access to an essential facility to a competitor in the downstream market. On the other hand, there is also the risk of false positives: the refusal to provide an essential input or to grant access to an essential facility may not be a sufficient test for identifying illegal conduct. 2 Case 27/76 United Brands v Commission [1976] ECR 207. Cf. van Bael & Bellis, Competition Law in the European Union, 4th ed. 2005, p. 947 (distinguishing United Brands as a sub-category of refusal-to-deal case from cases in which the dominant firm refuses to deal with a competitor in a downstream market). 2

3 2. Please state the statutory provisions or legal basis (including any relevant guidelines or formal guidance) for your agency to address a refusal to deal. Are there separate provisions for specific forms of refusal (e.g., IP licensing, essential facilities, margin squeeze)? Refusal-to-deal cases are dealt with under unilateral conduct rules. In many jurisdictions, it will be for the agencies and the courts to decide whether and if so under which conditions the general rules on unilateral conduct apply to a refusal to deal. U.S. law provides an example of such a law. Section 2 of the Sherman Act leaves it to practice to decide whether and under which conditions a refusal to deal has to be considered an act of monopolization or an attempt of monopolization. 3 Under EU law, a refusal to deal has to be dealt with under the general prohibition on abuse of market dominance. However, Article 82 of the EC Treaty 4 provides more guidance than U.S. law by containing a non-exhaustive list of examples of such an abuse. Yet, this list does not contain an explicit rule on refusal to deal. In individual cases of a refusal to deal, practices may refer to different examples contained in Art. 82. Most importantly, the European Court of Justice (ECJ) based its decision in Magill on Art. 82(b) EC, by arguing that the refusal to license the copyright in the listings of TV programs to an independent publisher prevented the appearance of a new product, a comprehensive weekly guide to television programmes, which the appellants did not offer and for which there was a potential consumer demand. 5 This new product requirement was explicitly derived from the formula of Art. 82(b) EC according to which an abuse may consist in limiting production, markets or technical development to the prejudice of consumers. Later on, the discussion arose whether the new product rule would have to be understood as a cumulative, i.e. indispensable requirement for a duty to license. In IMS Health, 6 The ECJ answered this question in the affirmative, however, without even mentioning Art. 82(b) EC. This judgment encouraged Microsoft to challenge the decision, in which the Commission had held that Micorsoft had violated Art. 82(b) EC by refusing to provide competitors in the market for work group server operating systems the interoperability information contained in the Windows program. Before the European Court of First Instance (CFI), Microsoft indeed argued that its competitors did not intend to provide a new product in the sense of the IMS Health holding. 7 However, the CFI recalled that the new product rule had been based by the ECJ on the example contained Art. 82(b) EC. 8 Thereby, the CFI was able to confirm the position of the Commission according to which the issue was not only whether competitors intended to offer a new product but that it was sufficient that Microsoft had limited technical development to the prejudice of consumers in the sense of Art. 82(b) by a refusal to allow follow-on innovation. 9 Still, Art. 82(b) EC should not be considered as the only legal basis for a prohibition of a refusal to deal under European law. It is to be recalled that in United Brands the ECJ did not even refer to the cases listed in Art. 82 EC, but immediately held that a dominant firm cannot stop supplying a long standing customer who abides by regular commercial practice, if the 3 See, in particular, Aspen Skiing v. Aspen Highlands Skiing, 472 U.S. 585 (1985); Verizon Communications Inc. v. Law Offices of Curtis V. Trinko, LLP, 540 U.S. 398 (2004). 4 These comments refer to the provisions of the EC Treaty still applicable at the time of the writing of these comments, although the Treaty on the Functioning of the European Union is expected come into force on 1 December Joined Cases C-241/91 P and C-242/91 P RTE and ITE v Commission (Magill) [1995] ECR I-743, para Case C-418/01 IMS Health [2004] ECR I-5039, para Case T-201/04 Microsoft v Commission [2007] ECR II-3601, paras Ibid., para Ibid., para

4 orders placed by that customer are in no way out of the ordinary. 10 Recognition of additional cases of a refusal to deal beyond the cases listed by Art. 82 EC is consistent with the wording of this provision since the list is not exhaustive. It has to be added that European practice could also rely on other examples of Art. 82 EC, especially on the prohibition on discrimination as contained in Art. 82(c) EC, in order to capture refusal-to-deal cases. Still in other jurisdiction, practice may even rely on rules that have specifically been adopted in view of refusal-to-deal cases. An example is provided by the Sec. 19(4) No. 4 German Act against Restraints of Competition, 11 which contains a specific rule on essential facilities. It is important to note that this rule is limited to networks and other infrastructure facilities. At the time of its adoption, in the late 1990s, the legislature carefully considered the scope of this rule and rejected its application to IP rights in particular. However, this has not prevented the German Federal Supreme Court (Bundesgerichtshof) from maintaining a duty to license of the holder of a patent that controlled access to a technological standard by relying on the German prohibition of discrimination as contained in Sec. 20(1) of the Act. 12 Hence, also German experience demonstrates that specific rules on refusal to deal should not be considered to be of an exclusionary character. Those very specific rules, just like the examples contained in Art. 82 EC, may be helpful in order to deal with some cases of a refusal to deal, whilst the general prohibition on anti-competitive unilateral conduct still provides a fall-back position for other cases. Rules implementing an essential facilities doctrine are also known in other jurisdictions. According to the Slovak Competition Act, also a right can be considered as an essential facility in the sense of its essential facilities provision. 13 Therefore, it seems that, in contrast to the German essential-facilities provision, the Slovak one would also apply to a refusal to license an essential intellectual property right. 3. Do the relevant provisions apply only to dominant firms or also to other firms? Some jurisdictions also prohibit unilateral conduct below the threshold of market dominance. In Germany, for instance, the prohibition on discrimination as contained in Sec. 20(1) Act against Restraints of Competition 14 also applies to firms that hold so-called relative market power. According to the second paragraph of Sec. 20, the first paragraph also applies to undertakings insofar as small or medium-sized enterprises as suppliers or purchasers of certain kinds of goods or commercial services depend on them in such a way that sufficient and reasonable possibilities of resorting to other undertakings do not exist. This rule has considerable practical importance in the retail business where small retailers may depend on access to certain brands for doing their business, although the brand as such does not provide market power, or inversely the producers of certain brands depend on access to large retail chains in order to have sufficient access to consumers. 10 Supra note 1, para This holding was recently confirmed in Joined Cases C-468/06 to C-478/06 Sot. Lélos kai Sia [2008] ECR I-7139; on this case see Josef Drexl, Healing with bananas How should Community competition law deal with restraints on parallel trade in pharmaceuticals? in TECHNOLOGY AND COMPETITION: CONTRIBUTIONS IN HONOUR OF HANNS ULLRICH 571 (Josef Drexl et al. eds 2009). 11 Gesetz gegen Wettbewerbsbeschränkungen (GWB). English translation available at: 12 Case KZR 40/02, (2004) Gewerblicher Rechtsschutz und Urheberrecht 966 Standard-Spundfass. An English translation of this decision is available in: 36 Journal of Intellectual Property and Competition Law (IIC) 742 (2005). 13 Art. 8(3) and (4) of the Act 136/2001 on the Protection of Competition. English translation available at: 14 Supra note 11. 4

5 4. Is a refusal to deal a civil/administrative and/or a criminal violation? If it is a criminal violation, does this apply to all forms of refusal to deal? In principle, the ban on refusal-to-deal cases are enforced in accordance with the general rules applicable to unilateral conduct. In contrast to the enforcement of the cartel prohibition, experience shows that private enforcement may play a major role in the field. Wherever the law makes a refusal to deal illegal, the law may recognize a corresponding private right of the potential customer (so-called petitioner ) to claim a duty to deal, which will be enforced by the private law courts. This is why, for instance, the abovementioned German rules on control of relative market power are mostly enforced by private parties themselves who seize private law courts with the goal of forcing the other party with relative market power to enter into a contractual relationship. Competition agencies may therefore largely abstain from enforcing such rules by imposing administrative fines. In general, private enforcement may even be more crucial in refusal-to-license cases. Here, the petitioner of an IP right may be tempted to simply use the right and may then be sued by the right-holder for infringement. The defendant may then rely on the violation of competition law as a defence to the infringement claim. Under EU competition law, such reliance on Art. 82 EC is known as the Euro defence. In a most recent case, the German Federal Supreme Court has brought precision to the requirements under which a firm can use another person s IP right and rely on German competition law as a defence to the IP infringement claim. 15 Another question is whether refusal-to-license cases are good cases for criminal sanctions. Since the assessment of the illegality of refusal-to-deal cases is most complex, for many jurisdictions the major problem in this regard will be that the legal provisions that apply are not sufficiently precise in the light of constitutional requirements for supporting criminal sanctions. IV. Experience 5. How many in-depth investigations (i.e., beyond a preliminary review) of a refusal to deal has your agency conducted during the past ten years (or use a different time frame if your records do not go back ten years)? See the answers provided by the competition agencies. 6. In how many refusal to deal cases did your agency find unlawful conduct during the past ten years? Please provide the number of cases concerning IP-licensing, essential facilities, margin squeeze, and all other types separately. For any case, in which your agency found unlawful behavior, please describe the anticompetitive effect and the circumstances that led to the finding. For administrative systems -- i.e., the agency issues its own decision (subject to judicial review) on the legality of the conduct -- please state the number of agency decisions 15 Bundesgerichtshof of 4 July 2009, Case KZR 29/06 Orange-Book-Standard. According to this judgment, the defendant is required to request a license before using the IP right. If the right-holder refuses to license, the defendant must still act according to a reasonable licensing contract and even pay reasonable licensing fees. 5

6 finding a violation, or settlements that were challenged in court and, of those, the number upheld and overturned. For judicial systems -- i.e., the agency challenges the conduct in court -- state the number of cases your agency has brought that resulted in a final court decision that the conduct violates the competition law or a settlement that includes relief. Please state whether any of these cases were brought using criminal antitrust authority. Please provide a short English summary of the leading refusal to deal cases (including IP licensing, essential facility, and margin squeeze) in your jurisdiction, and, if available, a link to the English translation, an executive summary, or press release. This question refers to the frequency of refusal-to-deal cases. Allthough these comments are not intended to report on the experience of any given jurisdiction, it still makes sense to consider the likelihood that refusal-to-deal cases will arise in the future in more general terms. Especially practice in experienced jurisdictions has proven to be very hesitant to intervene too easily and too frequently in refusal-to-deal cases. The reasons for such hesitation are very obvious. A ban on a refusal to deal does not prohibit a specific form of business action. It rather imposes a duty to act in a particular way. One may even argue that, when challenging a refusal to deal, the agencies and courts have to act very much like regulatory agencies, which sometimes will even have to fix the price for the good or service provided by the dominant firm. The agencies and courts may even be more hesitant to order a duty to license for not interfering with the exclusivity of IPRs as the very essence of such rights. Accordingly, in Magill, the ECJ held that a duty to license can only be affirmed in in exceptional circumstances. 16 But how do we have to understand the term of exceptional circumstances in this context? Is this meant to be a statement on the frequency of intervention or one on how to define the substantive threshold of intervention? In this regard, it is important to note that in the future we may well experience more cases involving a refusal to license. Already during the last few years, practice in the EU demonstrated that most important cases on refusal to deal were IP-related. Such cases may become even more frequent, when many markets, most importantly in the information technology (IT) sector, are increasingly based on standardized technologies. Access to such technology and, hence, to the IP rights that control the standard will be indispensable for competitors for entering the relevant market. Therefore, the question of whether unilateral conduct rules may be used as a legal basis for a duty to license may well become an almost daily issue for competition agencies in many jurisdictions rather than remaining just an exceptional phenomenon. 7. Does your jurisdiction allow private parties to challenge a refusal to deal in court? If yes, please provide a short description of representative examples of these cases. If known, indicate the number (or an estimate) of private cases. As already highlighted in the answer to Question 4, private cases may especially arise under Art. 82 EC. Within the EU, cases on a refusal to deal often reach the ECJ under the referral procedure of Art. 234 EC from private law courts who have before them cases of a Euro defence, like in refusal-to-license cases, 17 or cases in which a dominant firm is sued for 16 Supra note 5, para This was the case in the IMS Health case, supra note 6. 6

7 damages or even for a claim to start or to continue supplies. 18 In jurisdictions which allow such private enforcement, private parties have a choice between private enforcement and complaining before competition agencies with the objective of triggering administrative intervention. 19 Experience in the EU shows that competition agencies are not necessarily better placed than private law courts to provide quick relief to the victims of a refusal to deal. In IMS Health, for instance, the President of the CFI suspended the interim decision of the Commission, holding that the case-law of the courts was not sufficiently clear to provide a prima facie case against IMS Health who had refused to license to the complainant. 20 Private law courts may therefore be better placed than administrative agencies to provide interim relief to the victims of a refusal to deal, for being able to device a preliminary legal opinion in even more complex cases that have not been before the courts so far. VI. Evaluation of an actual refusal to deal 8. What are your jurisdiction s criteria for evaluating the legality of refusals to deal? You may wish to address the following points in your response. a. What are the competitive concerns regarding a refusal to deal? Must the practice exclude or threaten to exclude a rival (or rivals) from the market, or all rivals? If only threatened exclusion is required, how is it determined? If neither actual nor threatened exclusion is required, what other harms are considered? In very general terms, the primary competitive concern consists in that a refusal to deal has an exclusionary effect on rivals, whether they are potential customers or not, and thereby produces a restrictive impact on competition and/or reduces consumer welfare. Because of this exclusionary effect, refusal to deal is usually considered to belong to the so-called exclusionary practices, as opposed to purely exploitative practices. This raises the question whether actual exclusion should be required. From a policy perspective, the answer should be no. Unilateral conduct rules have a preventive function. They are designed to inform dominant firms on how they have to behave in the relevant market. Accordingly, firms have to assess whether they are allowed to act or not when they design their future business methods. Therefore, the only question can be whether the refusal to license has the potential effect of foreclosing markets and not whether a specific refusal to deal later on actually leads to the exclusion of one or several rivals from the market. Such predictions, of course, require an economic theory of harm. This is where the economics on market foreclosure has to come into play. One of these theories of harm is the essential facilities doctrine, which is based on the idea that a dominant firm that controls an essential facility (such as an infrastructure ferry harbours, airports, railway tracks) should 18 See more recently Joined Cases C-268/06 to C-278/06 Sot. Lélos kai Sia [2008] ECR I-7139 (private law claims of Greek pharmaceutical wholesalers against GlaxoSmithKline who tried to undermine parallel exports to other Member States by restricting supplies). 19 Both roads were taken in IMS Health; see Commission Interim Decision, Case COMP D3/ NDC Health/IMS Health, 20 Order of the President of the CFI of 26 October 2001, Case T-184/01 R IMS Health v Commission, ocjo&numaff=t-184/01&datefs=&datefe=&nomusuel=&domaine=&mots=&resmax=100. 7

8 not be allowed to exclude competition from the downstream market by refusing to grant access to the essential facility. However, refusal-to-deal cases should not always be equated with a leveraging theory, in which a dominant firm tries to extend market power from an upstream to a downstream market. Especially the ECJ has to be criticized for applying such an approach to some refusal-to-license cases. It may have been correct to apply a leveraging theory in the Magill case, 21 where it was possible to distinguish between the upstream market for the information on the TV programs (or maybe even the broadcasting operation itself) and the downstream market for TV guides. However, the situation in IMS Health was very different. 22 There, the petitioner was in need of using the copyrighted brick-structure of the dominant firm for entering the market for the service of providing data on the retail sales of drugs to the pharmaceutical companies after the 1860-brick structure, basically a map of Germany subdivided in 1860 sectors, had emerged as the standard used in the whole industry. In this case the cause of market dominance is to be found in the market for the service of data collection where the brick structure was used as a standardized tool. To distinguish here between an upstream market for the IP right and a downstream service market does not only appear highly artificial. More importantly, the leveraging theory, according to which the copyright holder harmed competition by leveraging market power from one market to another, does not really capture the core of the theory of harm needed for the assessment of the case. Rather, it would be more appropriate to state that there was only one relevant market, namely the service market, on which the copyright holder has been dominant all along and from which he was able to exclude any competitor by relying on the copyright controlling a standard that was essential for doing business in the market. Another issue is whether the potential of excluding some competitors would be sufficient or whether there must be complete elimination of competition by exclusion of all rivals. In the light of the general goal of consumer welfare, the latter might seem to provide the appropriate answer, since, as some would argue, competition law should not protect competitors but competition. As long as there is a sufficient number of competitors that keep prices low, consumers will not suffer. Yet this view needs to be rejected. Even discrimination against some competitors may be considered a specific aspect which may tip the balance to illegality. Whereas discrimination as such should not be considered as anti-competitive under unilateral conduct rules, even in the very cautious Trinko decision of 2004, 23 the U.S. Supreme Court held that a refusal to deal may only be considered illegal where a dominant firm stops supplying an existing customer or when it discriminates between different customers. In the above-cited German Standard-Spundfass case, 24 the German Federal Supreme Court affirmed a violation of German competition law in a situation where the holder of a patent on the technical specifications of a drum used in the chemical industry had agreed to grant licenses for free to all other German drum manufacturers who had participated in the standard-setting procedure of the German chemical industries for such drums but refused to license to an Italian manufacturer who had not participated in this procedure. For affirming a violation of competition law, the Court did not even consider whether the Italian competitor would have intended to provide higher quality drums, which was unlikely, given the fact that the Italian competitor was in need of using the same technology. Nor did the court argue that the exclusion of the Italian competitor had lowered price competition, which was equally unlikely, given the considerable number of other producers active in the market. This case may be taken as evidence that, under German law, competitors enjoy equal protection even if consumers do not suffer any specific harm. 21 Supra note See supra note Supra note Supra note 12. 8

9 Such a policy may be considered by many as economically unsound. However, the question of how many competitors are needed in order to drive prices down, in the sense of static efficiency, is already difficult to answer. Moreover, the question of whether specific conduct harms competition should not only be assessed through the lenses of neo-classical price theory but should also take into account institutional perspectives. Granting equal access to all competitors to a market does not only correspond to the normative criterion of equal treatment. It may also create the better institutional arrangement for promoting entrepreneurship and enhancing the willingness of individuals to take economic risk by starting their own business operations. The concept of competition in such a system is very much based on the notion of openness of the market guaranteed to all market participants and not just on consumer welfare defined through the criterion of static efficiency and pre-defined consumer interests. This concept of openness of markets may well be the better and more sustainable basis for an efficient, i.e. welfare enhancing competition law. Such an approach may also be preferable in view of enhancing dynamic competition. In a scenario of standardized technology, for instance, it may be better to have multiple players who compete with each other for follow-on innovation than just a limited number of players that are needed to keep prices low for consumers. Innovation processes are often much more driven by diversity of ideas, which requires that the most innovative minds have access to the market. b. Must consumer harm be demonstrated? Must the harm be actual or may it be just likely, potential, or some other degree of proof? Especially economists often argue that in order to distinguish properly between anticompetitive conduct and competition on the merits, there should be a requirement of a showing of consumer harm. 25 This also seems a widely spread view in the U.S. In the recent decision in Rambus v. FTC, a so-called patent hold-up case, in which the FTC had ordered Rambus to license its DRAM technology to chip manufacturers, the U.S. Court of Appeals for the District of Columbia (D.C. Circuit) held that a violation of Section 2 Sherman Act would require a showing of consumer harm. The Court thereby also clarified that only the final consumer could be considered as a consumer in the sense of the consumer harm criterion. Hence, a mere showing that chip producers would have to pay excessive prices was not held to be sufficient proof of harm to competition. 26 Under European law, the ECJ has never accepted such a consumer harm approach. Even according to most recent decisions of the ECJ 27 and of the CFI, 28 it was held that Art. 82 EC also protects consumers indirectly by protecting the competitive structure and that therefore a showing of actual harm to consumer would not be required. 29 In its GlaxoSmithKline decision, however, the CFI suddenly seemed to prefer a different approach to applying Art. 81 EC on restrictive agreements in requiring that in order to affirm a restraint of competition in the sense of Art. 81(1) EC actual harm to consumers needs to be demonstrated. The Court justified this view in the light of the overall goal of competition law to promote consumer welfare. On appeal, however, the ECJ corrected this holding by clarifying that EU competition law does also protect the structure of the market and that 25 See also the Economic Advisory Group on Competition Policy (EAGCP), An Economic Approach to Article 82, July 2005, 26 Rambus Inc. v. FTC (D.C. Cir. 2008), pdf. See also the critical comments by Josef Drexl, Deceptive Conduct in the Patent World A Case for US Antitrust and EU Competition Law, in: PATENTS AND TECHNOLOGICAL PROGRESS IN A GLOBALIZED WORLD, LIBER AMICORUM JOSEPH STRAUS 137, 143 et seq. (Wolrad Prinz zu Waldeck et al. eds 2009). 27 Case C-95/04 P British Airways v Commission [2007] ECR I-2331, para Case T-201/04 Microsoft v Commission [2007] ECR II-3601, para Case T-168/01 GlaxoSmithKline v Commission [2006] ECR II-2969, para

10 therefore it is not necessary that final consumers be deprived of the advantages of effective competition in terms of supply or price. 30 The European rejection of the consumer-harm approach is reminiscent of the structureconduct-performance paradigm which is nowadays quite rightly rejected by modern economics. However, the European approach to the criteria for a restraint of competition should not be accused of being based on a structural approach. The wording used by the Courts is explained by the specific set of facts underlying the early Continental Can judgment of the ECJ, in which it was held that ex-art. 86 EEC Treaty (Art. 82 EC) would also protect consumers in an indirect way by prohibiting a merger of an already dominant firm with a competing firm for harming the structure of competition. 31 Whereas in this decision, the Court had to deal with a merger case affecting the structure of competition, and thereby laid the foundation for the recognition the category of structural abuses as part of the Art. 82 EC, the wording ( structure ) later made it over to other abuses, including exclusionary practices and refusal-to-deal cases. It is worthwhile to note that in the Italian versions of the Continental Can decision and the successor decisions of the recent past, the structure wording has never been used. Instead, the Court in the Italian versions preferred to use the word of effective competition (concorrenza effettiva), which highlights the central argument of this case-law. Already in Continental Can, the Court relied on ex-art. 3(f) EEC Treaty (Art. 3(1)(g) EC) which proclaims the guarantee of undistorted competition in the internal market as one of the goals of EC law. From this, it is clear that the immediate goal of EC competition law is the protection of competition or the competitive process as such, whereas the protection of consumer interests is only an intermediary goal. The question remains, however, whether the European approach can be considered good economic policy. Indeed, European law may protect consumers even much better by protecting competition as such and without additionally requiring a showing of harm to consumers. Indeed, the consumer-harm approach suffers from a number of flaws. Two arguments seem most important: first, the consumer-harm approach increases the risk of false negatives by making it more difficult to prove a violation of competition law. 32 Secondly, there is also the need to predict whether a specific conduct is illegal when a dominant firm, for instance, refuses to deal with a rival. Economists would mostly assess consumer harm in terms of price and output. However, in many markets, technological progress has become increasingly important also for consumers. Whether a rival who is excluded from the market by a refusal to deal and who therefore will not be able to compete with the dominant firm for better products or follow-on innovation will come up with innovation that will be preferred by consumers, however, is simply a development that cannot be predicted easily or at all. Whether product innovations can convince consumers to move from the incumbent to the rival is a matter for the competitive market to decide. This is what Friedrich August von Hayek meant by competition as a discovery procedure. 33 According to this approach, it is better to protect the process of competition instead of assessing the effects of a conduct on competition by an attempt to predict future effects on the interest of consumers which can, after all, only be identified by effective competition. There is also an option to the consumer-harm approach. A restraint of competition especially in applying unilateral conduct rules should be assessed in the light of foreclosure effects. 30 Joined Cases C-501/06, C-513/06, C-515/06 and C-519/06 GlaxoSmithKline v Commission [2009] ECR I- 0000, para. 63; also citing Case C-8/08 T-Mobile Netherlands [2009] ECR I-0000, paras. 38 et seq. 31 Case C-6/72 Euroemballage and Continental Can v Commission [1973] ECR 215, para This is highlighted by the fact that in recent years it was for alleged infringers of competition law to argue before the European Courts that proof of actual harm to consumers should be required. 33 Friedrich-August von Hayek, Competition as a Discovery Procedure, in NEW STUDIES IN PHILOSOPHY, POLITICS AND ECONOMICS AND THE HISTORY OF IDEAS (Friedrich August von Hayek ed. 1978), p

11 However, in its Guidance Paper on Art. 82 EC, the Commission seems to go in the direction of the consumer-harm approach by arguing that it will focus its enforcement priorities on those types of conduct that are most harmful to consumers. 34 Yet this does not have to prove that the Commission will now completely switch to a consumer harm approach. In principle, it maintains the previous position in line with the case-law of the courts that a violation has to be proven by anticompetitive foreclosure 35 and provides a serious of considerations for assessing such foreclosure effects. 36 Still, the Commission seems to require likely consumer harm in addition to foreclosure effects and, in this regard, deserves to be criticized against the backdrop of the analysis presented above. At least the Commission clarifies that also direct purchasers, even when they are not final consumers, are to be considered consumers in this sense. In comparison to the situation in the U.S., this certainly mitigates the European approach. c. Does intent play a role, and if so what role and how is it demonstrated? In general, competition law should not penalize intent but only real harm or threat to competition. However, in practice, agencies and courts should and usually do take into account the strategies of dominant firms, including their intentions, when assessing the legality of their business conduct. In this regard, where investigation by the competition agencies produces evidence that a dominant firm has developed a strategy which it itself considers as one of exclusionary conduct, such evidence can support an economic-based theory of harm in the sense that the conduct indeed had the potential of foreclosing markets to actual or potential competitors. d. Are refusals to deal evaluated differently if there is a history of dealing between the parties? Is a prior course of dealing between the parties a requirement for finding liability? In the U.S., after the Trinko holding, 37 it looks very unlikely that the Supreme Court would ever apply Section 2 Sherman Act to an initial refusal to deal, whereas termination cases may still fall under Section 2 Sherman Act. In Europe, Article 82 EC has also been applied to cases of initial refusal to deal. 38 This is sound policy. To limit a violation of competition law to termination cases could even turn out to be counterproductive. Under such a rule, dominant firms would be better advised to never enter into a contractual relationship with rivals. Yet distinguishing between termination cases and initial refusals may make sense in the framework of assessing whether the refusal is efficient. Prior dealing indicates that the dominant firm itself was at least at one point in time of the opinion that this made economic sense. It is therefore good policy, like according to the Guidance Paper of the European Commission, to require the dominant firm to demonstrate why the dealing should no longer be considered efficient. 39 e. Are refusals to deal evaluated differently if the dominant firm has had a course of dealing with firms that are not rivals or potential rivals? Thus, if a firm sells its 34 Communication from the Commission Guidance on the Commission s enforcement priorities in applying Article 82 of the EC Treaty to abusive exclusionary conduct to dominant firms, 9 February 2009, para. 5, available at: 35 Ibid., para Ibid., para Supra note See, in particular, the ECJ decision in Magill, supra note 5, and IMS Health, supra note Guidance Paper, supra note 34, para

12 product to everyone except its main rival, is that relevant to whether the refusal is unlawful? Such discrimination may be another indication that the refusal to deal aims to exclude the rival from the market. Yet, also in this regard, there must be a convincing theory of harm which makes the refusal to deal likely to harm competition. Discrimination practiced by a dominant firm as such is no proof that there is no pro-competitive explanation of such conduct. Therefore, also in such cases, a more thorough economic analysis of the facts and the potential effects on competition needs to be undertaken. 9. Does your jurisdiction recognize a distinct offense of refusing to provide access to essential facilities? Your response need not include any offenses that arise from sector-specific regulatory provisions rather than the competition laws. If so, how does your jurisdiction define essential facilities? Under what conditions has a refusal to deal involving an essential facility been found unlawful? Please provide examples and the factors that led to the finding. As mentioned above, 40 some national laws have specific provisions on essential facilities with different scope of application. Other jurisdictions may recognize a specific essential facilities doctrine in applying the general unilateral conduct provisions. In the abovementioned Trinko judgment, the U.S. Supreme Court has explicitly refrained from either accepting or rejecting the essential facilities doctrine. Nevertheless, it is equally clear that this court would be very reluctant to accept such a doctrine in the future. In the EU, the ECJ is generally held to have developed a European essential facilities doctrine in a series of cases. These cases mostly relate to the refusal to license IP rights. 41 In the Bronner case, the ECJ relied on the Magill case-law on refusal to license without having to clarify whether the threshold for non-ip cases could possibly be lower. Since the Court held that access to the newspaper distribution system in Austria was not indispensable for the competitor to enter the market for daily newspapers, it was possible to reject a duty to deal for lack of an essential facility. 42 The indispensability criterion, which is essential for the application of Art. 82 EC to refusal-to-deal cases was developed in the very first essential facilities case in the Community, namely the Commercial Solvents case, in which an integrated firm terminated supply of essential raw materials in order to exclude competitors from a downstream market. 43 In this case, the ECJ highlighted in particular that ex-art. 86 EEC Treaty (Art. 82 EC) would have to apply if the refusal risks elimination all competition on the downstream market. 44 This requirement later made it to the case-law on refusal to license. 45 Whatever criteria are chosen for an essential facilities doctrine, an important caveat needs to be made. The mere existence of an essential facility does not provide a conclusive theory of harm. This plays in two directions. First, to the extent that the essential facilities doctrine relies on a leveraging theory, one has to keep in mind that there may also be refusal-to-deal cases which present an abuse but cannot be explained by a leveraging theory. 46 Second, the mere fact that a competitor is in need of access to an essential input controlled by a dominant 40 On Question 2, above. 41 Magill (supra note 5); IMS Health (supra note 6); see also the CFI decision in Microsoft (supra note 7). 42 Case C-7/97 Bronner [1998] ECR I-7791, paras 41 et seq. 43 Case 6/73 Istituto Chemioterapico Italiano and Commercial Solvents v Commission [1974] ECR Ibid., para Joined Cases C-241/91 P and C-242/91 P RTE and ITE v Commission (Magill) [1995] ECR I-743, para See also the answer on Question ##, above. 12

13 firm and that the refusal to deal (or to allow access to this input) does not by itself justify intervention. 47 A full assessment of the effects of the refusal to deal on the relevant market is certainly required. This assessment would have to take into account the negative impact of a duty to deal of the dominant firm s initial and subsequent incentives to invest in the provision of the essential facilities and in maintaining and improving it under such duty to deal. Equally, other justifications should be accepted such as limited capacity or technical compatibility and quality concerns. 9. Does the analysis differ if the refusal involves intellectual property? If so, please explain. In the U.S., there is no case-law on the level of the Supreme Court that would deal with a duty to license. However, in the light of the Trinko case, which uses clear language as to the importance of a monopoly for creating incentives for innovation, it may well be even likely than for regular refusal-to-deal cases that the Court would accept a duty to license IPRs. In the EU, the standard for refusal to license is defined by the ECJ judgment in IMS Health. This decision defined a cumulative three-factor test, which is actually a four-factor test if one takes into account that the bottom line of the test is that use of the relevant IP right has to be indispensable for being able to enter the relevant market. In IMS Health, the ECJ held: It is clear from that case-law that, in order for the refusal by an undertaking which owns a copyright to give access to a product or service indispensable for carrying on a particular business to be treated as abusive, it is sufficient that three cumulative conditions be satisfied, namely, that that refusal is preventing the emergence of a new product for which there is a potential consumer demand, that it is unjustified and such as to exclude any competition on a secondary market. 48 This decision was meant to clarify an earlier dispute as to whether the first requirement of the prevention of the emergence of a new product (so-called new product rule ) has to be considered a cumulative requirement, meaning that a refusal to license can only be accepted if the refusal leads to the prevention of a new product on the downstream market. The decision seems to affirm this question, which would amount to a narrow reading of the earlier Magill decision in this regard. Yet, on the new product rule, IMS Health does not solve all problems: First, it is hard to understand why the court, on the one hand, affirms the cumulative approach whereas, on the other hand, it also holds that the fulfillment of said cumulative requirements is sufficient (not necessary ) to show an abuse. This still brings up the question whether the decision leaves some room for accepting alternative tests under which a refusal to license can be considered an abuse. Such flexibility may also be derived from the fact that the new product rule was developed by the ECJ in Magill with reference to the wording of ex-art. 86(b) EC Treaty (Art. 82(b) EC), which only provides one possible example of an abuse. This is why the Commission, in its arguments before the CFI in the Microsoft case, referred to the wording of sufficient in IMS Health for supporting its view that IMS Health did not provide an exhaustive list of criteria under which a refusal to license can be considered abusive. 49 In Microsoft, the CFI later on seemed to make use of such flexibility by holding that according 47 There are a few cases on the Court of Appeals level that demonstrate a general reluctance to decide in favour of a duty to license; see, for instance, Data General Corp. v. Grumman System Support Corp., 36 F.3d (1st Cir. 1994). 48 Case C-418/01 IMS Health [2004] ECR I-5039, para Case T-201/04 Microsoft v Commission [2007] ECR II-3601, para